The kenja from niyaton

author nameやcommitter nameが特殊なエンコーディングを使用していると変換がうまくいかない

例えば以下のようなコミットをうまく変換できない．

apache/ant@f6bdb17

rename convertor -> converter

rename convertor.py to converter.py

Failed to setup kenja behind proxy

Thanks for the awesome tool!
I tried to install kenja with PC behind proxy and faced following problem.

$ python setup.py install
kenja/lib/java/java-parser.jar does not exist. Do you want to download it?[y/n]
y
md5 hash of kenja/lib/java/java-parser.jar is incorrect! remove it and tryagain.

This is because download_parser(self) method invokes urllib.urlretrieve() that does not consider proxy configuration.

    def download_parser(self):
        urllib.urlretrieve(self.parser_location, self.parser_path)

This can be modified by changing code as follows, since urllib2.urlopen() works based on http_proxy and https_proxy environment variables.

    def download_parser(self):
        with open(self.parser_path, 'wb') as f:
            f.write(urllib2.urlopen(self.parser_location).read())
            f.close()

Also I faced the similar problem on executing copy_csharp_parser() method. In this case I commented out the line and avoid the problem, since I only targets java source code in this time.

add 'git submodule init' to install.sh

submodule of git should be initialized when parent git repository is cloned from a remote.

Problem of maximum file name length.

Maximum length of the file name depends on file system.
However, Kenja make a file which have over 255 characters name in Historage(git).
We should consider how to avoid this problem.

Moving feature of recording syntax trees files to option.

When kenja converts historage, kenja makes syntax trees files into syntax_trees dir.
However, these files are temporal files.
These files are never used for mining Historage.

I think the feature storing syntax trees should be an option feature (for DEBUG).

Kenja cannot convert the history of ArgoUML to the Historage.

According to my friend, he cannot convert the history of ArgoUML to the Historage.
He got the git repository of the ArgoUML by using "git svn clone" command.
The location of ArgoUML SVN repository is following URL.
http://argouml.tigris.org/svn/argouml/

According to him, the name of a branch of the imported git repository contains @ and it raise some exception in a function of GitPython library.

メソッド抽出リファクタリングの検出で不思議な出力をする

jEditの履歴において，

抽出元メソッド：FtpVFS._createInputStream(View,Buffer,String,boolean)
抽出後のメソッド：getFtpClient
類似度：0.323943662

というメソッド抽出リファクタリングの候補が検出されるが，
_createInputStream(View, Buffer, String, boolean)というメソッドは
検出したコミット間のbefore側には存在せず，
_createInputStream(View,String,boolean)というメソッドのみが存在する．
これは，kenjaが勝手にリネームを含めたメソッド抽出の検出を試みているためであり，
また，抽出元のメソッド名をafter側から取得しているのが原因である．

メソッド抽出リファクタリングの検出時にリネームも考慮するのかどうかを検討した上でBug Fixを実施する．